Set the reasoning effort for a conversation

In NeuralVault 2.0 a chip in the chat footer opens a slider from Low to NV Ultra; the chip is absent when the app cannot confirm the current model supports it.

Last checked against the app on · View as Markdown

Steps

  1. Open a chat with a model that supports reasoning effort. Look in the footer of the message box for a chip with a gauge icon and a tier name, such as High.
  2. Press the chip. A panel headed "Reasoning effort" opens with a slider running from "Faster" to "Deeper". You can also scroll the mouse wheel over the chip to step through the tiers.
  3. Choose a tier. The tiers are Low, Medium, High, XHigh, Max and NV Ultra. Sending no effort at all behaves the same as High.
  4. Close the panel with Escape, by clicking outside it or by pressing the chip again. It stays open while you move the slider.

How to tell it worked

The chip shows the tier you chose. Its hover text reads "Reasoning effort for this conversation".

If it does not work

There is no effort chip.
It is left out, not greyed, whenever the app cannot confirm the current model supports it. It is only offered for Anthropic, OpenRouter and NeuralVault-routed models, so an Ollama model never shows it.
A model rejects XHigh.
Only newer models accept that tier. An older one may reject it. Choose High or Max instead.
NV Ultra is slow and costs more.
It runs the turn at Max, then a second full pass reviews and revises the answer, so it takes roughly double the time and token spend of Max. The tier carries a "Higher spending" badge.